A Middle Dorset dwelling in Trinity Bay, Newfoundland Une habitation du Dorsétien moyen dans la baie de Trinité, Terre-Neuve

Abstract

This paper describes a Middle Dorset dwelling in Trinity Bay Newfoundland. The dwelling features an extended flagstone pavement and a well defined lateral cooking area. The vast amount of refuse associated with the dwelling, as well as the radiocarbon dates ranging between 1880 and 1300 B.P., suggest that the house was occupied for a long period of time. The occupation of the dwelling was intense and complex and stratigraphic evidence points to at least five occupational or use episodes of the dwelling site.

Résumé

Cet article décrit une structure d'habitation du Dorsétien moyen dans la baie de Trinité à Terre-Neuve. L'habitation présente des éléments architecturaux particuliers dont un long dallage de pierres plates et une structure de combustion latérale soigneusement aménagée. La grande quantité de rebuts associée à l'habitation ainsi que les datations au radiocarbone s'étalant entre 1880 et 1300 A.A. suggèrent une installation de longue durée. La stratigraphie révèle une occupation intense et complexe comptant au moins cinq épisodes d'occupation ou d'utilisation de l'espace habité.

Introduction

Despite nearly eight decades of archaeological research on the island of Newfoundland, our knowledge of Palaeoeskimo architecture is still rather limited. Other than the well documented semi-subterranean Dorset dwellings at Port au Choix (Eastaugh 2002 and this volume; Harp 1976; Renouf 1986, 1987, 1991, 1992, 1993; Renouf and Murray 1999), there are but a few descriptions of Palaeoeskimo architecture on the island (Fogt 1998; Robbins 1985). The lack of architectural data is easily explained in that dwelling features must first be found. Unlike the Arctic context where structural remains are usually laying before our eyes, habitation structures in lower latitudes are in most instances deeply buried and therefore invisible on the surface. In such a context, finding a habitation structure involves a strong component of luck which in turn certainly affects the size of the Newfoundland sample. The fact that there is stratigraphy also means that we have to deal with a sedimentary envelope that can make the excavation of a house quite complex. At the same time, this sediment can yield evidence relating to the history of occupation of the dwelling.

For a long time, the Port au Choix semi-subterranean dwelling type has been recognized as the standard model in Newfoundland Palaeoeskimo architecture (Cox 1978; Harp 1976; Jordan 1986; Maxwell 1985). The most common house form at Port au Choix is a semi-subterranean dwelling featuring raised rear and/or lateral platforms surrounding a central depression, without pavement, which may or may not contain an axial feature (Eastaugh 2002; Harp 1976; Renouf 1986, 1987, 1991, 1992, 1993; Renouf and Murray 1999). Contrasting Middle Dorset dwellings in Newfoundland from those in Labrador, Cox (1978: 113) also notes that Newfoundland Dorset houses lack formal paved mid-passage features. However, as new data comes in (Erwin this volume; Fogt 1998; Hartery and Rast this volume), we can see the emergence of not only diversity in Newfoundland Palaeoeskimo architecture but formal flagstone pavements have now been revealed in Middle Dorset dwellings at Cape Ray (Fogt 1998) and in Trinity Bay (LeBlanc 1997; Robbins 1985).

In this context, the object of this paper is to introduce a Middle Dorset dwelling from Trinity Bay featuring a long and extremely well-preserved flagstone pavement. The discussion will focus first on the morphological description of the structure. Possible scenarios related to the history of occupation of the site will be discussed next. Finally, a close examination of post-depositional processes will shed new light on the interpretation of the dwelling.

Site description

The Dildo Island site is located at the bottom of Trinity Bay on the southeast coast of Newfoundland, approximately 2 km offshore from the community of Dildo (Figure 1). The island has an elongated shape and measures about 700 m north to south and 150 m east to west at its widest point. The northern portion of the island is high, covered by a coniferous forest and not easily accessible because of high cliffs. The island gradually slopes down to a grassy meadow which covers its entire south side. The southern part of Dildo Island is easily accessible from all directions as it is surrounded by cobble beaches. On the southwest side, Barry's Cove is well protected and provides the best access to the island. A fresh water spring runs from north to south through the island.

Fieldwork conducted on Dildo Island between 1995 and 1999 (Gilbert 1996; LeBlanc 1997, 1998, 1999) yielded evidence of 5,000 years of human occupation. The most recent use was a cod hatchery that operated in Barry's Cove in the late 1800s. The remains of Maritime Archaic Indians (5000-3200 B.P.) and Recent Indians (2000 B.P. to European contact) were found confined to the forested portion of the island. In Barry's Cove itself, two Palaeoeskimo cultures were also present: a small Groswater (2800-2100 B.P.) component and an extensive Middle Dorset (1900-1200 B.P.) occupation featuring the remains of two Dorset dwellings on the east side of the cove. Figure 2 shows Barry's Cove and the location of the dwellings.

Figure 1

Map of Newfoundland showing location of Dildo Island

House 1

House 1 will not be discussed at length in this paper, as it has been described elsewhere (LeBlanc 1997). However, certain features are worth mentioning as they contrast significantly with House 2. House 1 is located on a grassy terrace approximately 7 m above sea level. Its interpretation is limited as it was only partially excavated (16 m2). The excavation revealed a line of three hearths, some of which bordered by upright slabs, running along a northeast-southwest axis. The contour of the dwelling could not be defined with certainty but a low ridge of fire-cracked rocks on one side may indicate its eastern limit. Notably, a large amount of fire-cracked rocks were scattered within the structure. The shallow stratigraphy suggests surface structure(s); perhaps associated with a warm season occupation.

House 2

House 2 lies approximately 20 m down slope to the southwest of House 1, at an elevation of 4 to 5 m above sea level. Prior to excavation, the area was covered by tall grass and the unsuspected presence of the dwelling was fortuitously revealed by a 2 m x 2 m test pit. Laying 45 cm below surface, a portion of a flagstone pavement was uncovered, as well as part of what was soon recognized as the cooking area of the dwelling. When the grass was cut in preparation for excavation, a peat bulge interrupted by a narrow central groove was revealed (Figure 3). This break aligned with the pavement was at that time hypothesized as marking the entrance of the dwelling (Figure 4). The floorplan on Figure 4 shows the outline of the peat bulge.

A depression on the eastern side of the pavement clearly indicates that the dwelling has been dug into the hillside and marks its eastern limit. The western portion of the house rests on natural leveled subsoil, ending at the edge of a rock cliff. On that account, the dwelling can be described as partially semi-subterranean. The southern limit of the dwelling also ends at a cliff, suggesting that part of the house could have eroded away. Towards the north, the flagstone pavement opens up to a large and thick midden deposit.

Excavation of 80 m² of House 2 yielded a long flagstone pavement, a well defined cooking area with the possibility of a second cooking area, less well defined, and a large midden. The house is rich in artifacts: 5,562 lithic artifacts (including 12 soapstone fragments and a mobile lamp stand) and 117,627 flakes. Stylistically the artifacts are characteristic of the Dorset Trinity Bay Complex (LeBlanc 1997, 2000; Robbins 1985, 1986) featuring distinctive polished and serrated endblades, tiny quartz crystal scrapers and microblades. Although organic remains were scarce, a few bark specimens and an unidentified whalebone object were uncovered laying directly on the sterile clay at the bottom of the excavation.

Figure 2

Southern part of Dildo Island. In the foreground, the rectangular outline in the grass shows the location of House 1. House 2 is located downhill, by the excavation stakes.

Figure 3

Dildo Island House 2, showing peat bulge and groove before excavation

Figure 4

Floorplan, Dildo Island House 2 and associated Midden

The most prominent feature of the dwelling is the extremely well-preserved linear flagstone pavement 8.6 m long and approximately 2 m wide, oriented in a southeast-northwest direction. At its northern end, the pavement becomes narrower (about 1 m) suggesting the possibility of an entrance. This was reinforced by the corresponding break in the peat deposit immediately above. Were this the case, the dwelling would have had a north-facing entrance, towards Barry's Cove. The pavement is carefully constructed with intricately fitted slabs. Good quality flagstone rocks are available conveniently from a cliff on the west side of the cove (Figure 5). Except for three large boulders on its eastern side, the flagstone pavement is edged neither by slabs nor boulders.

No hearth features are integrated within the pavement area. Instead, adjacent and parallel to the pavement's northwest end, a cooking feature (Cooking Area 1 in Figure 6) is defined by a semi-circular arrangement of slanted rocks and slabs. It measures 1.93 m long by 1.60 m wide and its bottom is lined with flat slabs. A thick and well-sorted deposit of small pebbles encrusted with fat was mainly confined to the feature and flanked at each end by upright slabs. The anthropogenic nature of the deposit was initially questioned — it could have resulted from a storm dumping episode — but as the excavation progressed it soon became obvious that the deposit had built up over time as a large number of artifacts, broken slabs and fat concretions were found throughout. Partitioning upright slabs found at different depths within the deposit also hint at how specific sections of the cooking feature had been used over time. This pebble matrix not only represents an ideal and flexible stratum to support upright slabs, it also drains easily and a spoiled surface can be conveniently refreshed by adding more pebbles. While the sides and rear of the cooking feature are well delineated, its front shows no real edge and the pebble deposit joins up and slightly overflows the main pavement of the dwelling. Figures 7 and 8 show cross-sections of the Cooking Area I.

A similar small pebble deposit also showed up in a north-south profile (Profile W174 in Figure 9) towards the southern part of the dwelling. However, its excavation did not provide the same evidence; there were no traces of fat, slabs or an outline to suggest that it was used for cooking.

To the rear of the cooking area and parallel to the main pavement, an arrangement of flat rocks and slabs extends south for about 2 m. In part, the feature rests directly on bedrock, and is slightly raised above the main pavement; reminiscent of a lateral sitting/sleeping platform. During excavation, the area in front of this possible platform consisted of an extremely black, greasy and compacted deposit essentially free of rocks. The nature of the deposit and its location between the platform and the main pavement suggests that this was an area of intensive domestic activity within the dwelling. On the other hand, it may also imply reoccupation of the dwelling and hence reuse of that specific area. Another greasy deposit on the opposite side of the cooking area could be interpreted similarly.

Figure 5

View of Barry's Cove showing location of House 2 and flagstone rock outcrop

Figure 6a

Dildo Island House 2: Cooking Area 1 during excavation. Note the pebble matrix, fat concretions and upright slabs.

Figure 6b

Cooking Area 1 after excavation. Note the slanted rocks and slabs at the rear of the structure.

Figure 7

Profile W176 showing limits of Cooking Area 1

Figure 8

Profile N271, East/West cross-section of Cooking Area 1. The depression on the east side was dug into the hillside.

There is no direct evidence of hearth or fire-related material (such as fire-cracked rocks) within the dwelling. Oil, probably seal oil, was likely the only fuel used for cooking and heating. Three lamp stands were found in the dwelling (Figure 4). By lamp stand, I am referring to flat slabs exhibiting the blubber-stained outline of the overflow from rectangular soapstone lamp. Two of these blubber-stained slabs (Lamp stands 1 and 3) were integral parts of the pavement; the third (Lamp stand 2) was loose and found laying in the pebble deposit of Cooking Area 1. The bottom fragment of a rectangular soapstone vessel found in the vicinity of Cooking Area 1 nicely matched the outline of the blubber-stained overflow (about 26 cm across) of all three slabs.

To the north, the pavement ends and opens to a large, deep and complex midden. The full extent of the midden is not known. A long linear depression clearly marks its eastern limit and the thick cultural deposit showing in the northern wall indicates that the midden continues beyond the excavated area. Although no articulated house remains are visible, the depth, size and the obvious effort devoted to excavate the midden into the hillside all argue that it was a previous house depression. The midden deposit appears to have formed over a long period of time as artifacts, slabs, rocks and boulders are found at different depths throughout. Nine radiocarbon dates from the midden suggest at least 560 years of use between 1880 and 1320 cal B.P. (Table 1). Notably, there was no evidence of roof structure in the entire excavated area.

History of occupation

At first glance, House 2 seems to have been built in a single episode. There are three arguments for this interpretation. The first is the high degree of integrity of the flagstone pavement, which runs for over 8 m without interruption. Secondly, the flagstone pavement and the bottom of the cooking area appear contemporaneous as they join and rest at the same level. Finally, three radiocarbon dates reinforce this interpretation: a date from 1350 cal B.P. was obtained from the cooking area itself, and dates of 1310 cal B.P. and 1300 cal B.P. from nearby on the flagstone pavement.

However, upon closer examination, the human occupation of the building site itself has a more complex history than the impressive remains of the dwelling would indicate. This complexity revealed in the stratigraphic profiles shows at least five construction or use episodes within the excavated area. A schematic representation (Figure 9) of the flagstone pavement along the stratigraphic profile W174 illustrates that: 1) the southern portion of the pavement is resting on sterile subsoil; 2) moving north, the pavement is running over a pre-existing cultural deposit (Deposit 1); 3) further north, the pavement extends slightly over an even deeper cultural deposit (the midden). The dotted lines show the two depressions corresponding with the deposits. The dotted lines in Figure 4 indicate the location of the same two depressions in relation to the floorplan.

Figure 9

Schematic representation of flagstone pavement along Profile W 174

The homogeneity of the cultural deposits (Deposit 1 and the idden) makes it difficult to identify and isolate chronologically specific occupational or use episodes. However, by establishing a series of relationships between the deposits and the flagstone pavement, we can propose a provisional sequence of occupation for the excavated area. Figure 10 illustrates this sequence.

Episode 1: Midden/House depression

The earliest episode is represented by the midden deposit. This deposit clearly pre-existed the construction of the dwelling pavement as illustrated by the slabs resting above part of it. If my interpretation of the midden as a previous house depression is correct, a likely scenario is that the house was abandoned and used as a midden for an extended period of time between 1880 and 1320 cal B.P.

Episode 2: Deposit 1

A second episode of use is represented by Deposit 1. As is the case for the midden deposit, Deposit 1 also accumulated over a previous depression (although shallower) located in the vicinity of Cooking Area 1. The outline of the depression is shown on the floorplan (Figure 4). Deposit 1 and the associated depression is an extremely complex event to interpret. Although stratigraphic evidence clearly shows that Deposit 1 pre-dates the pavement, its temporal relationship with the midden is difficult to establish. On one hand, conceivably Deposit 1 could be an extension of or result from the overflow of the midden deposit; on the other hand, the whole area could also have been filled and leveled in order to set the pavement and cooking feature in this part of the dwelling.

Episode 3: Tent ring

A third episode is represented by a circular formation of large boulders in the midden area (see floorplan in Figure 4). The boulder formation is reminiscent of a tent ring and the area may have been chosen as a tent site towards the end of the midden formation. The boulders were laying at the same depth, in the upper portion of the midden deposit, at approximately the same stratigraphic level as the flagstone pavement, which may suggest the contemporaneity of the two features. Unfortunately, radiocarbon dates obtained in the vicinity of the tent ring structure do not allow testing this hypothesis as they were all collected in the lower portion of the midden deposit.

Episode 4: Pavement construction

A fourth episode corresponds to the construction of the flagstone pavement. As mentioned above, that event clearly post-dated both the formation of the midden and Deposit 1. Its contemporaneity with the tent ring is difficult to establish with certainty but remains possible.

Episode 5: Peat cutting

Episode five refers to a peat cutting event which occurred after site formation. This event will be discussed in detail below.

Radiocarbon results

Eighteen radiocarbon dates were obtained from House 2 and the midden (Table 1). Table 1 shows a steady and tight sequence of occupation between 1880 and 1300 cal B.P. placing the dwelling site well within the Middle Dorset range of occupation on the island of Newfoundland. A cluster of dates in the 1600s could indicate a more intensive use during that period. With the exception of one date (1660 ± 60 cal B.P.) collected near the platform, the earliest dates were all obtained from the midden area, reinforcing the hypothesis of its early age. Dates associated with the flagstone pavement fall between 1530 cal B.P. and 1300 cal B.P. and one date of 1310 cal B.P. was obtained from the cooking area Figure 11 presents a summary of the dates in relation to the midden, Deposit 1 and pavement. It is important to note that the radiocarbon samples obtained in the vicinity of Deposit 1 were only collected above the flagstone pavement. This decision was made in order to preserve the integrity of the pavement. If we were to return to the site, additional information could be drawn from the comparison of dates below and above the flagstone pavement.

Post-depositional processes

The last issue addressed in this paper brings us back to the grooved peat bulge interpreted earlier as possibly marking the entrance of the dwelling. Peat blocks were commonly used as building elements in Palaeoeskimo architecture (e.g., Tanfield site in Maxwell 1985: 157) and it is easy to assume that peat was used for the construction of one of the walls of House 2. Were this the case, we should expect to find peat confined to specific areas around the dwelling. Profile W174 (Figure 9) is informative, showing that the peat layer is not localized but covers a vast portion of the cultural deposit. It is also clear from the profile that peat deposition occurred on top of, or after the cultural deposit had formed and therefore postdates the occupation of the dwelling. Moreover, the homogeneous texture of the peat deposit, the absence of structural remnant of peat blocks and the absence of mineral soil at the interface of the peat and the underlying cultural deposit (commonly associated with imported peat blocks, Schweger, pers. comm. 2002) all argue for in situ peat deposition.

In fact, peat formation at this location is not surprising considering the dwelling sits in a depression at the bottom of a hill. The accumulation of water combined with the nutrient rich environment of the cultural deposits provided an ideal substratum upon which vegetation developed and accumulated leading to the formation of a thick peat layer over the site.

Table 1

Radiocarbon dates of Dildo Island House 2 and associated midden

Note: The radiocarbon dates are calibrated at 1 sigma according to the curves in Stuiver et al. (1998).

Interestingly, Profile W 174 also features, over the midden, a clear interruption or cut in the peat layer. The peat was obviously cut at this location. In reference to the floor plan (N272/W173), this interruption is continuous with the peat bulge outline which was earlier tentatively interpreted as being a dwelling's wall. This hypothesis can now be dismissed. Exactly who excavated the peat is not known. What we can say is that it happened sometime after the Dorset people abandoned the site, long enough after for a thick layer of peat to form. Perhaps, when the cod hatchery was in activity in Barry's Cove in the late 1800s, people may have taken advantage of this convenient source of peat.

Figure 10

Provisional sequence of occupation of House 2 and associated midden

Figure 11

Spatial distribution of radiocarbon dates within House 2 and associated midden

Discussion

The work on Dildo Island revealed the remains of a Middle Dorset dwelling featuring an extensive and well preserved flagstone pavement and a well defined lateral cooking area. From a Newfoundland perspective, the architectural features described here are unusual in two respects. First, except for the Cape Ray (Fogt 1998) and Stock Cove (Robbins 1985) sites, flagstone pavements have rarely been reported at Middle Dorset sites in Newfoundland. In fact, such pavements appear to become more common during the Late Dorset period, by which time the island of Newfoundland had been abandoned by the Dorset people. For instance, in a Late Dorset context, such features have been described at Okak-3 on the Labrador coast (Cox this volume and 1978), at the Snowdrift site at Port Refuge, High Arctic (McGhee 1981) and at DIA-4, Ungava (Plumet 1976). Might we see in the Dildo Island pavement a hint of what was to come in the Late Dorset period?

The second uncommon trait is the lateral cooking features; in most instances, in Palaeoeskimo architecture, cooking features are integrated within the central pavement of the dwelling. At the 63Ø1-111-56 site, in the district of Skjøldungen, Greenland, Jensen (1996: 153) mentions, without assigning any function, a "large compartment filled with small pebbles" at one end of the mid-passage of a Dorset dwelling; at the Porden Point site, McGhee (1981: 20) describes a lateral box-like structure at one end of the mid-passage of an Independence II dwelling; finally at the DIA.4-A (JfEl-4 - Diana Island) site, associated with Dorset artifacts, Plumet (1976: 199 and 203) refers to a well constructed lateral combustion feature at the southeast end of the pavement area. In fact, House 2 on Dildo Island is reminiscent in many respects of the DIA.4-A house: both exhibit an extended flagstone pavement and a lateral cooking feature. Both also share a similar complexity and intensity of the occupation (see Badgley 1980; Plumet and Badgley 1980).

House 2 on Dildo Island is a complex dwelling. Most certainly, the dwelling bears Dorset architectural traits, but what is unique is their idiosyncratic combination. This singularity is unequivocally a product of the Dorset culture but we have to keep in mind that it is also the resultant of 600 years of occupation and modifications. Further research on Dorset architecture may cast a different light on the special nature of the Dildo Island dwelling.